metadata
license: creativeml-openrail-m
language:
- en
tags:
- text-to-image
- midjourney
- stable-diffusion
- disco-diffusion
- art
inference: true
Paint Journey V2 is Paint Journey V1 fine-tuned on 768x768 oil paintings by Midjourney, Open Journey, and Disco Diffusion
To apply the oil paint effect, begin the prompt with ((oil painting)). For starters, try using a dimension of 768x768 for any subject, 512x768 for portraits, and 1152x768 for landscapes.
To further improve the resolution and reduce noise in generated images, especially when using the model for img2img, use Paint Journey VAE in combination with Checkpoint.
As an example, a Automatic1111's WebUI user can add both of these files to the ./stable-diffusion-webui/models/Stable-diffusion folder.
Portrait Examples
Landscape Examples
Outerspace Examples
Prompts from top to bottom:
((oil painting)), portrait of a cute attractive model, ((round pupil)), smirk, ((beautiful eyes)), high resolution, uhd wallpaper
Negative prompt: low-res blurry fog mutated body ((duplicate)) ((double head)) ((long neck)) ((extra eyes)) error ((extra fingers)) long face long body ((watermark)) signature
Steps: 20, Sampler: Euler a, CFG scale: 7.5, Seed: 874776980, Size: 512x768, Model hash: 26fd2f6f, Batch size: 4, Batch pos: 0
((oil painting)), portrait of a handsome young man, small smile, athlete, ((beautiful eyes)), high resolution, uhd wallpaper
Negative prompt: low-res blurry fog mutated body naked ((duplicate)) ((double head)) ((long neck)) ((extra eyes)) error ((extra fingers)) long face long body ((watermark)) signature
Steps: 20, Sampler: Euler a, CFG scale: 7.5, Seed: 2389251230, Size: 512x768, Model hash: 26fd2f6f, Batch size: 4, Batch pos: 0
((oil painting)), lighthouse near the sea, high resolution wallpaper
Negative prompt: low-res blurry fog watermark
Steps: 20, Sampler: Euler, CFG scale: 7.5, Seed: 353450032, Size: 768x768, Model hash: 26fd2f6f, Batch size: 4, Batch pos: 0
((oil painting)), a lush valley with mountains in the background, high resolution, uhd wallpaper
Negative prompt: low-res blurry fog watermark
Steps: 20, Sampler: Euler, CFG scale: 7.5, Seed: 65984537, Size: 1056x768, Model hash: 26fd2f6f, Batch size: 4, Batch pos: 0
((oil painting)), a forest, moon, dark, night, high resolution, uhd wallpaper
Negative prompt: low-res blurry fog watermark
Steps: 20, Sampler: Euler, CFG scale: 7.5, Seed: 4247689747, Size: 1056x768, Model hash: 26fd2f6f, Batch size: 4, Batch pos: 0
((oil painting)), stunning landscape in an alien world, high resolution, uhd wallpaper
Negative prompt: low-res blurry fog watermark
Steps: 20, Sampler: Euler, CFG scale: 7.5, Seed: 3693048118, Size: 1056x768, Model hash: 26fd2f6f, Batch size: 4, Batch pos: 0
((oil painting)), the milky way, high resolution, uhd wallpaper
Negative prompt: low-res blurry fog watermark
Steps: 20, Sampler: Euler, CFG scale: 7.5, Seed: 3904195208, Size: 1056x768, Model hash: 26fd2f6f, Batch size: 4, Batch pos: 0
((oil painting)), the universe, high resolution wallpaper
Negative prompt: low-res blurry fog watermark
Steps: 20, Sampler: Euler, CFG scale: 7.5, Seed: 432451301, Size: 768x768, Model hash: 26fd2f6f, Batch size: 4, Batch pos: 0
((oil painting)), a futuristic space station, lighting, sci-fi, dry paint, high resolution, uhd wallpaper
Negative prompt: low-res blurry ((fog)) watermark
Steps: 20, Sampler: Euler, CFG scale: 7.5, Seed: 4009095562, Size: 1056x768, Model hash: 26fd2f6f, Batch size: 4, Batch pos: 0
Automatic1111's WebUI
git clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.git
Download paint_journey_v2.ckpt to the ./stable-diffusion-webui/models/Stable-diffusion folder. Run webui-user.bat.
Diffusers
pip install --upgrade diffusers
from diffusers import StableDiffusionPipeline
import torch
pipe = StableDiffusionPipeline.from_pretrained("FredZhang7/paint-journey-v2")
pipe = pipe.to("cuda")
prompt = "((oil painting)), a boat sailing, night sky, high resolution, uhd wallpaper"
image = pipe(prompt).images[0]
image.save("./result.png")
Safety Checker
The official stable diffusion safety checker uses up 1.22GB VRAM. Recommend using Google Safesearch Mini (99MB) to save 1.12GB VRAM.